Nardelli Mimma, Valenza Gaetano, Greco Alberto, Lanata Antonio, Scilingo Enzo Pasquale
Annu Int Conf IEEE Eng Med Biol Soc. 2015;2015:6110-3. doi: 10.1109/EMBC.2015.7319786.
This study reports on the recognition of different arousal levels, elicited by affective sounds, performed using estimates of autonomic nervous system dynamics. Specifically, as a part of the circumplex model of affect, arousal levels were recognized by properly combining information gathered from standard and nonlinear analysis of heartbeat dynamics, which was derived from the electrocardiogram (ECG). Affective sounds were gathered from the International Affective Digitized Sound System and grouped into four different levels of arousal. A group of 27 healthy volunteers underwent such elicitation while ECG signals were continuously recorded. Results showed that a quadratic discriminant classifier, as applied implementing a leave-one-subject-out procedure, achieved a recognition accuracy of 84.26%. Moreover, this study confirms the crucial role of heartbeat nonlinear dynamics for emotion recognition, hereby estimated through lagged Poincare plots.
本研究报告了利用自主神经系统动力学估计对由情感声音引发的不同唤醒水平的识别。具体而言，作为情感环形模型的一部分，通过适当组合从心跳动力学的标准分析和非线性分析中收集的信息来识别唤醒水平，心跳动力学信息来自心电图（ECG）。情感声音取自国际情感数字化声音系统，并被分为四个不同的唤醒水平。27名健康志愿者在连续记录心电图信号的同时接受了这种诱发。结果表明，应用二次判别分类器并采用留一法程序，识别准确率达到了84.26%。此外，本研究证实了心跳非线性动力学在情绪识别中的关键作用，在此通过滞后庞加莱图进行估计。
